Metabolic engineering of living organisms is still in its infancy with respect to industrial applications, despite the fact that genetic engineering has now been feasible for more than a decade. To a large extent, this may be due to the disappointing outcome of many of the attempts so far to improve strain performance. There are at least two reasons for the negative outcome of the attempts to increase metabolic fluxes:
One is that the genetic engineer tends to overlook the subtlety of control and regulation of cellular metabolism. The expression of enzymes that are expected to be rate limiting are increased 10 to 100 fold, e.g. by placing the gene on a high copy number plasmid. Or, a branching flux in a pathway is eliminated by deleting a gene. Quite often, this will have secondary effects on the metabolism, for instance by lowering metabolite concentrations that are essential to other parts of the cellular metabolism (e.g. processes that are essential to the growth of the organism) and the net result may be that the overall performance of the cell with respect to the desired product is decreased. Instead, it is necessary to tune the expression of the relevant gene around the normal expression level and determine the optimal expression level, for instance as the level that maximizes or minimizes the flux.
The second reason for the negative outcome lies in the rate limiting concept itself: both metabolic control theory (Kacser and Burns, 1973) and experimental determinations of control by individual steps in a pathway (Schaaff et al., 1989; Jensen et al., 1993) have shown that reaction steps which were expected to be rate limiting with respect to a particular flux, turned out to have no or very little control over the flux. Instead, the control and regulation of the cellular metabolism turned out to be distributed over several enzymes in a pathway, and it may be necessary to enhance the expression of several enzymes in order to obtain a higher flux.
According to metabolic control theory, the total flux control exerted by all the enzymes in a pathway, should always sum up to 1. Therefore, after one enzyme concentration has been optimized, the flux control will have shifted to another enzyme(s), and it may then be useful to perform additional rounds of enzyme optimization in order to increase the flux further.
In summary, flux optimization requires 1) fine-tuning of enzyme concentration rather than many fold overexpression and often 2) optimization of the level of several enzymes in a pathway rather than looking for the rate limiting step.
There are now many systems available that allow one to increase the gene expression more than 1000 fold and/or to turn on gene expression at a particular time point during a fermentation process (e.g. using temperature inducible systems). With respect to tuning the steady state gene expression in the fermenter, to say 150% or 70% of the normal expression level, it becomes more difficult. In principle, one could use a lac-type promoter in front of the gene of interest, and then add a certain amount of an inducer of the lac system, for instance IPTG (isopropyl-β-D-thiogalactoside), or use a temperature sensitive system at the correct temperature. These possibilities are often not practical for large scale industrial applications. The alternative is to use a promoter that has exactly the right strength. However, such promoters are seldom available, and furthermore one needs a range of promoter activities in order to optimize the expression of the gene in the first place, see below.
During the past two decades, much work has been done to define and optimize the consensus sequences of microorganisms. In many prokaryotes, one often finds two more or less conserved DNA sequences at approximately position −10 and −35 relative to the start site for transcription, TATAAT and TTGACA, respectively, with aproximately 17 basepairs between the two. The dogma in this field is that, by including these elements, the resulting promoters would tend to become strong. Indeed, promoter up mutations, which are relatively rare events, usually results in a better match to the above consensus sequences, while down mutations results in a poorer match to the consensus sequences or a less optimal distances between these. In addition, when random DNA sequences are cloned in place of one of the two consensus sequences, the strength of the resulting promoters usually correlate with the degree of homology to the consensus sequences.
In principle, modulation of the strength of promoters could then be achieved by basepair changes in the consensus sequences or by changes in the length of the spacer between these. But the impact of such changes on the promoter strength will tend to be large (see example 1 of this invention), and it is therefore not feasible to achieve small steps of strength modulation through base pair changes in the consensus sequences.
While the length of the spacer separating the two consensus sequences is known to play an important role for the strength of a promoter, the sequence of the spacers between the consensus sequences has usually been considered to be of little importance for the strength of the promoters, and attempts to identify additional consensus sequences in the spacer region through mutagenesis have indeed been unsuccessful. So far, nobody has attempted to randomize the spacer, while keeping the consensus sequences and the spacer length relatively constant.
Numerous experiments have been carried out in order to define and optimize the consensus sequences of microorganisms, including experiments where at least one of the consensus sequences was being randomized. In some of these experiments a part of the nucleotides surrounding the consensus sequences was also randomized in order to allow for the generation of promoters with length of spacers different from 17 bp, and/or in order to find possible new consensus motifs around the consensus sequences. The chances that this will generate an efficient promoter is very small and a selection must be applied in order to find those rare cases where the homology to the consensus sequence is sufficiently high to result in even a weak promoter.